How would you describe them?
maxxing the ability to run different LLMs on the smallest devices possible
